It turns the sources you follow into a personal radio station. The app reads from real sources, distills them into clear narration, and plays them back as a continuous stream.
The product is built around on-device AI, so the experience is not dependent on metered cloud usage. The reason you can press Play is because the AI is running on your iPhone.
Features include:
- Gravitas Radio for feed-based listening
- Research Assistant for source-grounded takeaways and drilldown
- Discover Assistant for finding curated articles and feeds
- Focus mode for prompting an entire source, like a full article, PDF, or EPUB, in one context-aware session
- imports for articles, feeds, PDFs, EPUBs, and more
Toward the UX of spatial computing: beyond controllers
Spatial computing is still in its early days. The hardware is here, but the interaction language is only starting to emerge. We don’t yet have the “mouse” or the “touchscreen gesture” equivalent for this new medium.
In building Gravitas Dark Matter for Apple Vision Pro, I wanted to see what would happen if I pushed past the familiar metaphors of controllers, joysticks, and handhelds — and designed around something simpler: the eye and a pinch.
Eye Pinch Targeting
What I call Eye Pinch Targeting combines two inputs that Vision Pro offers natively:
Eye gaze: natural head + eye motion to aim.
Pinch: a subtle, controller-free gesture to confirm.
The result surprised me. It didn’t feel like a hack or a compromise. It felt precise — almost inevitable. The dexterity of a gentle pinch gave me more confidence than any floating joystick UI could.
It’s not a virtual gun. It’s not a VR controller disguised as a wand. It’s something native to the medium.
Modes of spatial play
The second experiment was seeing how play changes depending on how much of the environment you keep:
Passthrough mode: your actual room remains visible, with targets anchored in physical space.
Apple Immersion: the curated environments Apple ships with Vision Pro.
Galaxy immersion: a full spatial takeover — black holes and celestial targets filling your field of view.
Each mode carried different affordances. Passthrough made play feel like AR. Apple’s environments gave a filmic layer. The galaxy mode created scale and immersion you simply can’t fake with a flat screen.
Hardcore stress testing
Performance was a hurdle. Cranking graphics to High at first made the experience unplayable. But after optimization passes, I was able to run a “hardcore” session: Arena Mode across the entire length of my home — living room, dining room, kitchen, hallway.
It was physical. You tread lightly to avoid throwing off tracking, but in the heat of it I spun quickly, weaving through space. I lost multiple rounds. It felt more like a sport than a demo. And unlike pre-canned VR attractions (like The VOID at Downtown Disney), this wasn’t on rails. It was reactive, competitive, and punishing.
Why this matters for UX
Spatial computing won’t scale if input feels clumsy or bolted-on. Eye Pinch Targeting showed me that precision and intuitiveness don’t need controllers. They can come from inputs already built into the hardware — if we design for them.
The broader takeaway: UX in spatial computing isn’t about recreating old metaphors. It’s about letting interaction feel native to the body and the environment.
Closing thoughts
Gravitas Dark Matter is still experimental. There are hitches, optimizations, and unanswered questions. But the exercise points toward something larger: that the UX of spatial computing may not look like controllers at all. It may look like gaze, pinch, and movement — gestures that feel inevitable once you experience them.
Gravitas Threads plays Reddit posts as an interactive, gamified experience. The extension aims to reduce scrolling prevalent in social networks by replacing the wall with an interactive 3D volume of posts. Users can play the simulation to receive recommendations based on heuristics such as upvotes, newness, top, hot, and so on. In addition, the user can pause the simulation to pick video and image posts by clicking on the interactive elements. The last 10 absorbed posts get displayed in a pane which are clickable links to the Reddit pages.
Gravitas Threads is available for free as a google chrome extension.
I built the app based off a gravitational formula I've been working on which diverges from Newtonian gravity and Kepler's Laws and finds an averaged gravitational pull from all bodies in the system. This provided an elegant and incredibly efficient real-time physics engine which is built on ThreeJs. The motion mimics planetary orbits and also insects swarming to a certain extent.
Gravitas Threads is unique because it is a real-time recommendation engine which is fast and free in contrast to AI which costs businesses untold, vast amounts of dollars to build and maintain significant models.
Hands-Free Reddit?
I built a Chrome extension that lets you browse Reddit without scrolling.
Instead of a 2D grid, Gravitas Threads presents posts in a 3D space, letting you explore dynamically using a physics-based recommendation system (GREP—Gravitas Recommendation Engine Protocol).
GIFs & videos autoplay and get fetched as needed.
No infinite scroll—just fluid, interactive browsing
Explore content like a real-time data visualization
Reddit posts become dynamic objects you can interact with
Why I Built This
Most social platforms force you to scroll endlessly. I wanted a system where posts are treated as dynamic objects, and you can explore them naturally. Also, I wanted to create a recommendation system which busts up the "rich get richer" dynamics of most algorithms.
We're excited to announce the launch of Gravitas Discover, the first vertical built on the Gravitas Recommendation Engine Protocol (GREP). This high-speed, physics-driven Shopify app is designed to optimize storefronts and enhance conversion rates for e-commerce merchants.
What is GREP?
The Gravitas Recommendation Engine Protocol (GREP) is an open-source, modular framework designed for scalable, high-performance recommendation systems. It offers a flexible, decentralized approach to data-driven decision-making, prioritizing speed, security, and adaptability.
With Gravitas Discover, Shopify merchants can leverage real-time data processing and physics-powered insights to refine their stores, improve customer engagement, and drive sales like never before.
The Gravitas Code is Open for Viewing
In the spirit of transparency and innovation, we’ve made the Gravitas code publicly available under a Non-Commercial & Educational License.
Why It Matters
1⃣ Lightning-fast performance – Optimized for Shopify stores to process recommendations in real-time.
2⃣ Decentralized & transparent – Open for viewing and improvement.
3⃣ Scalability & modularity – The GREP framework isn’t limited to e-commerce; it’s designed to support a variety of industries and applications.
What’s Next?
This is just the beginning. GREP is built to expand, with new verticals and applications in the pipeline. If you're a developer, entrepreneur, or researcher, we invite you to dive into the code, explore the architecture, and join the conversation.
I'm excited to share Gravitas, a source-available recommendation engine inspired by the mechanics of gravitational physics. By leveraging Gravitational Averaging (GA) and Biomimetic Gravitational Averaging (BGA), Gravitas provides a customizable, physics-based alternative to traditional recommendation systems.
The system visualizes "recommendations" as a dynamic simulation of interactions between smaller Dynamic Nodes (DNs) and larger Primary Mass Nodes (PMNs). This approach creates a gamified and intuitive way to understand and refine recommendation logic, offering:
Controlled Chaos: A unique balance between deterministic and probabilistic outcomes.
User Control: Adjustable sliders for prioritizing precision or speed in recommendations.
Customizable Configurations: From e-commerce product searches to dating apps, Gravitas can simulate a variety of recommendation environments.
Gravitas is source-available under a non-commercial license, meaning it’s free to use, modify, and experiment with for personal or educational purposes. Commercial inquiries are welcome.
If you've ever wondered what it would look like to "art direct" recommendations—or explore a recommendation engine built on physics-inspired mechanics—check it out!
Would love to hear your thoughts, feedback, or ideas for applying Gravitas!
It turns the sources you follow into a personal radio station. The app reads from real sources, distills them into clear narration, and plays them back as a continuous stream.
The product is built around on-device AI, so the experience is not dependent on metered cloud usage. The reason you can press Play is because the AI is running on your iPhone.
Features include:
- Gravitas Radio for feed-based listening
- Research Assistant for source-grounded takeaways and drilldown
- Discover Assistant for finding curated articles and feeds
- Focus mode for prompting an entire source, like a full article, PDF, or EPUB, in one context-aware session
- imports for articles, feeds, PDFs, EPUBs, and more
- high-quality narration with PocketTTS
- a growing Audiobook mode for long-form sources
Demo: https://youtu.be/xag31FaFg2U?si=07xgPOKPMkCfRFTM
TestFlight: https://testflight.apple.com/join/pqmbzjMa
reply